基于稀疏编码的手指心电信号身份识别算法

研究了心电(ECG)信号在身份识别中的应用,提出了基于过完备字典下稀疏编码的手指心电身份识别认证算法.在预处理阶段,对ECG信号进行预处理消噪,去除心电信号里的噪声、基线漂移和心率变异的干扰.在特征提取阶段,提取单周期心电信号构成特征向量并构建字典模型,用核奇异值分解(KSVD)训练成冗余字典,然后对每一部分特征向量进行稀疏编码,实现在该字典上的稀疏表示.在分类识别阶段,利用得到的稀疏系数矩阵构建特征模板向量作为特征参数.通过欧氏距离匹配输出个体身份信息,实现个体身份识别认证.通过两个手指心电信号数据库对该算法进行了性能测试,获得了较高的识别率.

Algorithm for finger-ECG signal identity recognition based on sparse coding

Application of ECG signal in identity recognition is studied and is proposed,sparse coding algorithm based on complete dictionary.In preprocessing stage,denoise noises in ECG like baseline wander and interference of heart rate variability.In feature extraction stage,extract single periodic signal to form feature vectors.Utilize the feature vectors to build the dictionary model and then be trained to be redundant dictionary by KSVD.Then every part of feature vectors is used for sparse coding,implement sparse representation in the dictionary.In classification stage,utilize sparse coefficient matrix to build characteristic template vector and use it as characteristic parameters.Then matched output individual identity information through Euclidean distance realize personal identity authentication.Performance test on the proposed algorithm is carried out through two fingers ECG database named CYBHi and surface ECG data and won high recognition rate.